Visão Geral

Este Curso Deep Learning and Development of Generative AI Models Fundamentals, analisa brevemente os conceitos de aprendizagem profunda e, em seguida, ensina o desenvolvimento de modelos generativos de IA. 

Pre-Requisitos

Os alunos deverão ter experiência prévia no desenvolvimento de modelos de Deep Learning, incluindo arquiteturas como Redes Neurais artificiais feed-forward, recorrentes e convolucionais. 

Materiais

Conteúdo Programatico

Review of Core Python Concepts (**if needed – depends on tool context**)

1. Anaconda Computing Environment
2. Importing and manipulating Data with Pandas
3. Exploratory Data Analysis with Pandas and Seaborn
4. NumPy ndarrays versus Pandas Dataframes

Overview of Machine Learning / Deep Learning

1. Developing predictive models with ML
2. How Deep Learning techniques have extended ML 
3. Use cases and models for ML and Deep Learning 

Hands on Introduction to Artificial Neural Networks (ANNs) and Deep Learning 

1. Components of Neural Network Architecture
2. Evaluate Neural Network Fit on a Known Function
3. Define and Monitor Convergence of a Neural Network
4. Evaluating Models
5. Scoring New Datasets with a Model

Hands on Deep Learning Model Construction for Prediction 

1. Preprocessing Tabular Datasets for Deep Learning Workflows
2. Data Validation Strategies
3. Architecture Modifications for Managing Over-fitting
4. Regularization Strategies
5. Deep Learning Classification Model example
6. Deep Learning Regression Model example 
7. Trustworthy AI Frameworks for this DL prediction context

Generative AI fundamentals:

1. Generating new content versus analyzing existing content
2. Example use cases: text, music, artwork, code generation
3. Ethics of generative AI

Sequential Generation with RNN

1. Recurrent neural networks overview
2. Preparing text data
3. Setting up training samples and outputs
4. Model training with batching
5. Generating text from a trained model
6. Pros and cons of sequential generation

Variational Autoencoders

1. What is an autoencoder?
2. Building a simple autoencoder from a fully connected layer
3. Sparse autoencoders
4. Deep convolutional autoencoders
5. Applications of autoencoders to image denoising
6. Sequential autoencoder
7. Variational autoencoders

Generative Adversarial Networks

1. Model stacking
2. Adversarial examples
3. Generational and discriminative networks
4. Building a generative adversarial network 

Transformer Architectures

1. The problems with recurrent architectures
2. Attention-based architectures
3. Positional encoding
4. The Transformer: attention is all you need
5. Time series classification using transformers

Overview of current popular large language models (LLM):

1. ChatGPT
2. DALL-E 2
3. Bing AI

Medium sized LLM on in your own environment:

1. tanford Alpaca
2. Facebook Llama
3. Transfer learning with your own data in these contexts